Summary: New research shows that people perceive AI systems as more creative when they observe not only the final product but also the creative process and the robot. In a series of controlled experiments with identical drawings, participants consistently rated the creative abilities as more creative than when they observed the process itself.
Interestingly, the physical appearance of the robot had little effect on these ratings, contradicting previous assumptions about design bias. These findings have important implications for how we design, evaluate, and interpret AI creativity, and potentially even how we assess human creativity.
Important facts:
- Understanding is essential: Creativity scores increased as viewers saw more action and robots.
- The shape of the robot is not relevant: There is no significant difference in perceived creativity between the two robot designs.
- Design effects: The presentations influence how we understand AI creativity, raising ethical and practical design questions.
Source: Aalto University
Why do people think AI systems are creative?
New research suggests that it depends on how much they view the creative process.
“AI is playing an increasingly important role in creative practice. Whether that means we should call it creative or not is another question,” says Nikki Pennen, the lead author of the study.
Pennanen researches AI systems at Aalto University and has a background in psychology. Together with Aalto and other researchers at the University of Helsinki, he conducted experiments to determine whether people perceive robots as more creative when they closely observe their creative process.
In the study, participants were initially asked to judge the robots’ creativity based entirely on still-life drawings they had created themselves. They were told that the robots were controlled by AI, but in reality, they were programmed to reproduce drawings created by researchers of an artist.
This illusion allowed people’s perception of creativity to be measured without the robots being creative. Otherwise, there would have been a lot of differences between the drawings.
Participants then assessed their drawing creativity. They not only saw the final product, but also watched a video of the drawing process (the lines are visible on the page, but the robot does not draw them).
In the final stage, participants judged the drawing when they could see all three elements: the final product, the process, and the robot that created the drawing.
The results showed that drawings were considered more creative because more elements of the creative process were revealed.
“The more people saw it, the more creative it was,” says Christian Gokelsberger, assistant professor of creative technologies at Aalto University and lead author of the study.
To my knowledge, we are the first to study the effects of looking at product, process, and producer separately and in a controlled manner, not only in the context of AI, but more broadly.
The power of perception
Understanding how people assess the creativity of robots or other artificial systems is important so that we can design them. However, it is not yet entirely clear which design choices are appropriate for this.
“But if we add elements to make the AI system appear more creative, when in reality the system works the same way, we can question whether it’s such a good idea.”
In some cases, this can be useful—for example, it can be a way to help people stay connected to a co-creative system. But in other contexts, it can give the misleading impression of the actual creation of an artificial system.
“Our findings help us resolve this conflict by giving us a better understanding of our human biases. This research makes them a little more transparent, which is also important from a user perspective, so we can understand how the design of the system affects our perception of it,” Guckelsberger said.
In addition to these social and design implications, the findings are also relevant to research on creative AI systems. If our judgment of creativity depends on how the system is presented, future studies should consider this factor.
Existing research needs to be re-evaluated based on these findings: comparing the creativity of different systems without taking into account differences in their presentation may have reached erroneous conclusions.
Another interesting question this research raises is what it tells us about ourselves. “Now that we’ve discovered this about people’s perception of AI creativity… does this also apply to people’s perceptions of others?” Guckelsberger asks.
Does shape matter?
The researchers also conducted experiments with two different robot designs. Their goal was to test whether people evaluate the creativity of a robot differently depending on its appearance. Previous research has already suggested a link between appearance and perceived creativity.
The team examined whether people experienced different levels of creativity when drawing still lifes with a robot with a stylized arm or a more mechanical drawing robot. Maintaining consistency in the drawings between the robot and the participants was a challenge.
“I think our biggest challenge was the physical robots. We worked a lot with the robots and the design process to make everything the same so we could make scientifically rigorous comparisons,” says Pennen.
The researchers were surprised that there was no significant difference in the ratings of the two robots. They plan to delve deeper into this conflicting result in the future, as well as into other factors that influence our perception of creativity.
“We are interested in further investigating what types of biases influence the evaluation of our creative and embodied AI systems and how these effects arise,” Pennanen said.
The findings should be confirmed for different artistic genres, as well as other forms of art and creative expression. To make it easier for others to replicate and build on their work, the researchers followed open science methods.
As artificial systems become increasingly common, understanding the factors that shape our perception of their creativity is crucial for effective design. It could also shed light on how we recognize creativity in humans.
About this AI and creativity research news
Author: Sarah Hudson
Source: Aalto University
Contact: Sarah Hudson – Aalto University
Image: The image is credited to StackZone Neuro
Original Research: Open access.
“Is AI truly creative? Turns out creativity is in the eye of the beholder” by Niki Pennanen et al. ACM Transactions on Human-Robot Interaction
Abstract
Is AI really creative? It turns out that creativity depends on the observer.
Although creative artificial intelligence (AI) is becoming an integral part of our lives, we know very little about what makes us call “creative” AI.
Building on previous theoretical and experimental work, we investigate how perceiving evidence of a creative process beyond the final product influences our assessment of a robot’s creativity.
We studied body morphology as a potential moderator of this relationship, leading to a 3 × 2 factorial design.
In two laboratory visual arts experiments, participants (N = 30 + 60) examined drawings produced by two physical robots of different shapes, while facing three levels of perceptual evidence: product, process, and producer.
The data supports that human appreciation for the creativity of robots increases significantly as more is revealed about the process of creation and ultimately the product itself, in addition to the producer.
We found no significant effects of avatar form, which contradicts current hypotheses and provides more detailed insights for future research.
The latter is also based on additional exploratory analyses that reveal factors that may influence assessments of creativity, including perceived friendliness toward the robot and participants’ experience with robotics and AI.
Our insights empirically validate existing design patterns, promote fairness and accuracy in system comparisons, and contribute to a better understanding of our relationship with creative AI and thus its acceptance in society.
